Presentation is loading. Please wait.

Presentation is loading. Please wait.

6-1 ©2006 Raj Jain www.rajjain.com Clustering Techniques  Goal: Partition into groups so the members of a group are as similar as possible and different.

Similar presentations


Presentation on theme: "6-1 ©2006 Raj Jain www.rajjain.com Clustering Techniques  Goal: Partition into groups so the members of a group are as similar as possible and different."— Presentation transcript:

1 6-1 ©2006 Raj Jain www.rajjain.com Clustering Techniques  Goal: Partition into groups so the members of a group are as similar as possible and different groups are as dissimilar as possible.  Statistically, the intragroup variance should be as small as possible, and inter-group variance should be as large as possible. Total Variance = Intra-group Variance + Inter-group Variance

2 6-2 ©2006 Raj Jain www.rajjain.com Clustering Techniques (Cont)  Nonhierarchical techniques: Start with an arbitrary set of k clusters, Move members until the intra-group variance is minimum.  Hierarchical Techniques:  Agglomerative: Start with n clusters and merge  Divisive: Start with one cluster and divide.  Two popular techniques:  Minimum spanning tree method (agglomerative)  Centroid method (Divisive)

3 6-3 ©2006 Raj Jain www.rajjain.com Minimum Spanning Tree-Clustering Method 1.Start with k = n clusters. 2.Find the centroid of the i th cluster, i=1, 2, …, k. 3.Compute the inter-cluster distance matrix. 4.Merge the the nearest clusters. 5.Repeat steps 2 through 4 until all components are part of one cluster.

4 6-4 ©2006 Raj Jain www.rajjain.com Minimum Spanning Tree Example  Step 1: Consider five clusters with ith cluster consisting solely of ith program.  Step 2: The centroids are {2, 4}, {3, 5}, {1, 6}, {4, 3}, and {5, 2}.

5 6-5 ©2006 Raj Jain www.rajjain.com Spanning Tree Example (Cont)  Step 3: The Euclidean distance is:  Step 4: Minimum inter-cluster distance =  2. Merge A+B, D+E.

6 6-6 ©2006 Raj Jain www.rajjain.com Spanning Tree Example (Cont)  Step 2: The centroid of cluster pair AB is {(2+3)  2, (4+5)  2}, that is, {2.5, 4.5}. Similarly, the centroid of pair DE is {4.5, 2.5}.

7 6-7 ©2006 Raj Jain www.rajjain.com Spanning Tree Example (Cont)  Step 3: The distance matrix is:  Step 4: Merge AB and C.  Step 2: The centroid of cluster ABC is {(2+3+1) ¥ 3, (4+5+6) ¥ 3}, that is, {2, 5}.

8 6-8 ©2006 Raj Jain www.rajjain.com Spanning Tree Example (Cont) Spanning Tree Example (Cont)  Step 3: The distance matrix is:  Step 4: Minimum distance is 12.5. Merge ABC and DE  Single Custer ABCDE


Download ppt "6-1 ©2006 Raj Jain www.rajjain.com Clustering Techniques  Goal: Partition into groups so the members of a group are as similar as possible and different."

Similar presentations


Ads by Google